Community Services Strategy: Survival analysis

Author

Sarah Lucas

Published

2025-01-29

Aims

This analysis aims to investigate the risk of death or subsequent readmission in patients who have had an emergency hospital admission. It will help us understand what factors influence the risk of death or readmission, including determining variability between ICB area. This could help identify where activity might be mitigated through appropriate community services provision.

Methodology

Three patient cohorts are included within this analysis:

  • Elderly emergency: those 75 or over with an emergency admission.

  • Frail: those 65 or over with an emergency admission and a frailty score over 5. The frailty score is calculated from ICD-10 diagnoses recorded for admissions during the previous 2 years and using the risk scores in Gilbert et al (2018), Development and validation of a Hospital Frailty Risk Score focusing on older people in acute care settings using electronic hospital records: an observational study. The Lancet, 391(10132), pp.1775-1782. (Note these patients are a subset for those in the elderly emergency cohort)

  • Falls: those 65 or over with an emergency admission related to a fall.

The End of life cohort are not included as by definition they die during their hospital admission.

Survival analysis has been conducted using data from 2022/23 for

  • Time to death (mortality)

  • Time to readmission (time to readmission from the first admission for all patients including some that will have died in the year post-admission)

  • Time to readmission- excluding those who died prior to readmission

Note: Frimley ICB has been removed from the analysis due to missing data.

Mortality

The time from when a patient first had an emergency admission and first entered the cohort until death was calculated. Those who died on the same day as their emergency admission when they first entered the cohort were excluded.

Summary of percentage of patients dying within 1 year (2022/23)

Elderly Emergency

Frail

Falls

Sex

female

158,781 (26.9%)

117,967 (26.9%)

19,996 (21.5%)

male

151,814 (31.7%)

118,018 (30.8%)

14,284 (30.9%)

Age

65-69

N/A

15,214 (16%)

885 (9.6%)

70-74

N/A

23,743 (19.3%)

1,884 (13%)

75-79

65,677 (20.1%)

36,450 (22.9%)

3,848 (16.9%)

80-84

74,262 (25.3%)

45,672 (28.3%)

6,219 (22%)

85-89

83,602 (32.7%)

54,655 (35.2%)

9,195 (27.9%)

90+

87,054 (45.1%)

60,251 (47%)

12,249 (39%)

Ethnicity

Asian/Asian British

6,521 (23.8%)

5,353 (22.2%)

484 (19.3%)

Black/Black British

3,228 (25.2%)

2,521 (24.4%)

229 (19.6%)

Mixed

660 (25%)

572 (24.9%)

64 (21.6%)

Other

3,126 (26.3%)

2,480 (24.6%)

315 (19.9%)

White British

257,148 (29.4%)

194,921 (29.3%)

28,789 (25.2%)

White Other

10,276 (28.1%)

7,922 (27.8%)

1,169 (23.1%)

Unknown

29,636 (28.4%)

22,216 (27.1%)

3,230 (22.4%)

IMD decile

1

26,468 (31.3%)

22,398 (29.6%)

2,898 (25.2%)

2

27,682 (30.9%)

22,692 (29.9%)

3,260 (26%)

3

28,332 (30.1%)

22,496 (29.2%)

3,210 (25.1%)

4

31,182 (30.2%)

23,778 (29.5%)

3,401 (25.3%)

5

32,184 (29.2%)

24,374 (29.1%)

3,488 (24.5%)

6

33,531 (28.8%)

24,890 (28.5%)

3,735 (24.7%)

7

34,270 (28.7%)

25,238 (28.6%)

3,737 (24.7%)

8

33,722 (28.2%)

24,869 (28.4%)

3,611 (23.9%)

9

32,720 (27.5%)

23,706 (27.5%)

3,666 (24.4%)

10

30,504 (27%)

21,544 (27%)

3,274 (23.1%)

Total

310,595 (29.1%)

235,985 (28.7%)

34,280 (24.6%)

Comparing time to death between the cohorts using Kaplan-meier plot (2022/23)

For those first admitted in 2022/23 we considered the time to death from when the person was first admitted. The most recent year 2023/2024 was not used as there has not be sufficient time for those that entered the cohort at the end of this year to have a full year of follow-up time.

The risk of death is highest in the frail cohort, followed by the elderly emergency cohort, with those in the falls cohort have a lowest risk of death.

Kaplan Meier survival plots of time to death (2022/23)

Summary

The risk of death is increased:

  • with age

  • in males

  • with increasing deprivation

  • in those who are White British, compared to all other those of other ethnicities

There is some variation between ICBs in the risk of death with the lowest risk seen in the North West London ICB area.


ICB

Deaths

No. at risk

Percentage

QRV: NHS North West London ICB

7,111

29,342

24.2

QUY: NHS Bristol, N S'set & S Gloucs ICB

4,839

17,961

26.9

QRL: NHS Hampshire And Isle Of Wight ICB

11,089

41,017

27.0

QMJ: NHS North Central London ICB

4,317

15,633

27.6

QHL: NHS Birmingham And Solihull ICB

6,637

24,002

27.7

QNC: NHS Staffordshire & Stoke-On-Tr ICB

7,241

26,098

27.7

QOX: NHS Bath & NE S'set, S'don & W ICB

5,065

18,290

27.7

QXU: NHS Surrey Heartlands ICB

6,163

22,254

27.7

QU9: NHS Bucks, Oxford & Berkshire W ICB

8,288

29,859

27.8

QK1: NHS Leics, Leic'shire & Rutland ICB

5,317

19,088

27.9

QVV: NHS Dorset ICB

5,914

21,182

27.9

QOC: NHS Shrops, Telford & Wrekin ICB

3,107

11,024

28.2

QJK: NHS Devon ICB

8,178

28,784

28.4

QKS: NHS Kent And Medway ICB

10,551

37,198

28.4

QKK: NHS South East London ICB

6,492

22,696

28.6

QR1: NHS Gloucestershire ICB

3,451

12,067

28.6

QWE: NHS South West London ICB

5,628

19,703

28.6

QMF: NHS North East London ICB

5,715

19,918

28.7

QSL: NHS Somerset ICB

4,231

14,692

28.8

QT6: NHS Cornwall & Isles Of Scilly ICB

3,078

10,670

28.8

QHM: NHS North East & North Cumbria ICB

20,584

70,998

29.0

QWU: NHS Coventry And Warwickshire ICB

5,646

19,453

29.0

QNX: NHS Sussex ICB

11,294

38,875

29.1

QHG: NHS Beds, Luton & Milton Keynes ICB

4,697

16,045

29.3

QUA: NHS Black Country ICB

7,696

26,260

29.3

QM7: NHS Hertfordshire & West Essex ICB

8,321

28,276

29.4

QYG: NHS Cheshire And Merseyside ICB

15,964

54,391

29.4

QJ2: NHS Derby And Derbyshire ICB

6,934

23,505

29.5

QF7: NHS South Yorkshire ICB

8,775

29,424

29.8

QE1: NHS Lancashire & South Cumbria ICB

11,140

37,233

29.9

QPM: NHS Northamptonshire ICB

4,213

14,111

29.9

QH8: NHS Mid And South Essex ICB

7,523

24,890

30.2

QJM: NHS Lincolnshire ICB

5,139

17,010

30.2

QJG: NHS Suffolk & North East Essex ICB

6,913

22,736

30.4

QOQ: NHS Humber And North Yorkshire ICB

11,800

38,865

30.4

QUE: NHS Cambs & Peterborough ICB

5,041

16,461

30.6

QGH: NHS Herefords & Worcestershire ICB

4,909

16,003

30.7

QOP: NHS Greater Manchester ICB

15,127

49,098

30.8

QT1: NHS Nott'ham & Nottinghamshire ICB

6,628

21,522

30.8

QWO: NHS West Yorkshire ICB

12,907

41,047

31.4

QMM: NHS Norfolk And Waveney ICB

7,598

23,897

31.8

Summary

The risk of death is increased:

  • with age

  • in males

  • with increasing deprivation

  • in those who are White British, compared to all other those of other ethnicities

There is some variation between ICBs in the risk of death, with many of the London ICB areas having a lower risk of death.


ICB

Deaths

No. at risk

Percentage

QT6: NHS Cornwall & Isles Of Scilly ICB

230

1,199

19.2

QMJ: NHS North Central London ICB

503

2,542

19.8

QRV: NHS North West London ICB

848

4,080

20.8

QOX: NHS Bath & NE S'set, S'don & W ICB

544

2,488

21.9

QOC: NHS Shrops, Telford & Wrekin ICB

286

1,294

22.1

QU9: NHS Bucks, Oxford & Berkshire W ICB

843

3,754

22.5

QKS: NHS Kent And Medway ICB

755

3,305

22.8

QWE: NHS South West London ICB

661

2,893

22.8

QNX: NHS Sussex ICB

1,254

5,485

22.9

QRL: NHS Hampshire And Isle Of Wight ICB

1,392

6,056

23.0

QNC: NHS Staffordshire & Stoke-On-Tr ICB

646

2,749

23.5

QXU: NHS Surrey Heartlands ICB

678

2,869

23.6

QKK: NHS South East London ICB

987

4,167

23.7

QVV: NHS Dorset ICB

635

2,659

23.9

QYG: NHS Cheshire And Merseyside ICB

1,609

6,724

23.9

QJK: NHS Devon ICB

770

3,206

24.0

QJ2: NHS Derby And Derbyshire ICB

594

2,467

24.1

QMF: NHS North East London ICB

821

3,392

24.2

QJM: NHS Lincolnshire ICB

508

2,085

24.4

QT1: NHS Nott'ham & Nottinghamshire ICB

648

2,648

24.5

QGH: NHS Herefords & Worcestershire ICB

532

2,160

24.6

QHM: NHS North East & North Cumbria ICB

2,203

8,901

24.8

QUE: NHS Cambs & Peterborough ICB

561

2,257

24.9

QSL: NHS Somerset ICB

467

1,859

25.1

QE1: NHS Lancashire & South Cumbria ICB

1,242

4,929

25.2

QJG: NHS Suffolk & North East Essex ICB

748

2,949

25.4

QM7: NHS Hertfordshire & West Essex ICB

983

3,867

25.4

QPM: NHS Northamptonshire ICB

468

1,844

25.4

QHL: NHS Birmingham And Solihull ICB

714

2,803

25.5

QK1: NHS Leics, Leic'shire & Rutland ICB

599

2,318

25.8

QWU: NHS Coventry And Warwickshire ICB

655

2,539

25.8

QOP: NHS Greater Manchester ICB

1,656

6,383

25.9

QHG: NHS Beds, Luton & Milton Keynes ICB

639

2,427

26.3

QOQ: NHS Humber And North Yorkshire ICB

1,165

4,415

26.4

QWO: NHS West Yorkshire ICB

1,470

5,489

26.8

QMM: NHS Norfolk And Waveney ICB

911

3,380

27.0

QR1: NHS Gloucestershire ICB

414

1,523

27.2

QF7: NHS South Yorkshire ICB

1,016

3,727

27.3

QUA: NHS Black Country ICB

966

3,525

27.4

QUY: NHS Bristol, N S'set & S Gloucs ICB

704

2,573

27.4

QH8: NHS Mid And South Essex ICB

1,046

3,649

28.7

Summary

The risk of death is increased:

  • with age

  • in males

  • with increasing deprivation

  • in those who are White British, compared to all other those of other ethnicities

There is some variation between ICBs in the risk of death, with the lowest risk of death seen in the North West London ICB area.


ICB

Deaths

No. at risk

Percentage

QRV: NHS North West London ICB

5,721

24,250

23.6

QOX: NHS Bath & NE S'set, S'don & W ICB

3,471

13,075

26.5

QNC: NHS Staffordshire & Stoke-On-Tr ICB

5,512

20,688

26.6

QXU: NHS Surrey Heartlands ICB

4,447

16,721

26.6

QVV: NHS Dorset ICB

4,065

15,111

26.9

QHL: NHS Birmingham And Solihull ICB

5,082

18,771

27.1

QUY: NHS Bristol, N S'set & S Gloucs ICB

3,654

13,473

27.1

QRL: NHS Hampshire And Isle Of Wight ICB

8,134

29,919

27.2

QU9: NHS Bucks, Oxford & Berkshire W ICB

6,232

22,657

27.5

QMJ: NHS North Central London ICB

3,329

12,015

27.7

QK1: NHS Leics, Leic'shire & Rutland ICB

4,007

14,401

27.8

QWE: NHS South West London ICB

4,383

15,728

27.9

QT6: NHS Cornwall & Isles Of Scilly ICB

2,101

7,494

28.0

QKK: NHS South East London ICB

5,318

18,941

28.1

QOC: NHS Shrops, Telford & Wrekin ICB

2,322

8,266

28.1

QSL: NHS Somerset ICB

3,041

10,841

28.1

QHM: NHS North East & North Cumbria ICB

16,130

57,065

28.3

QR1: NHS Gloucestershire ICB

2,432

8,537

28.5

QNX: NHS Sussex ICB

8,312

29,057

28.6

QJK: NHS Devon ICB

5,813

20,270

28.7

QMF: NHS North East London ICB

4,610

16,078

28.7

QYG: NHS Cheshire And Merseyside ICB

12,764

44,434

28.7

QKS: NHS Kent And Medway ICB

7,140

24,797

28.8

QUA: NHS Black Country ICB

6,048

20,990

28.8

QHG: NHS Beds, Luton & Milton Keynes ICB

3,616

12,450

29.0

QF7: NHS South Yorkshire ICB

6,710

22,908

29.3

QJM: NHS Lincolnshire ICB

3,874

13,220

29.3

QM7: NHS Hertfordshire & West Essex ICB

6,252

21,310

29.3

QWU: NHS Coventry And Warwickshire ICB

4,424

15,123

29.3

QJ2: NHS Derby And Derbyshire ICB

5,246

17,797

29.5

QH8: NHS Mid And South Essex ICB

5,806

19,418

29.9

QE1: NHS Lancashire & South Cumbria ICB

8,728

28,924

30.2

QOP: NHS Greater Manchester ICB

11,979

39,638

30.2

QPM: NHS Northamptonshire ICB

3,411

11,307

30.2

QOQ: NHS Humber And North Yorkshire ICB

8,940

29,499

30.3

QWO: NHS West Yorkshire ICB

10,110

33,171

30.5

QT1: NHS Nott'ham & Nottinghamshire ICB

5,140

16,821

30.6

QUE: NHS Cambs & Peterborough ICB

3,899

12,688

30.7

QGH: NHS Herefords & Worcestershire ICB

3,652

11,825

30.9

QJG: NHS Suffolk & North East Essex ICB

5,161

16,517

31.2

QMM: NHS Norfolk And Waveney ICB

5,522

17,639

31.3

Multivariate survival analysis of time to death (2022/23)

For this analysis we have taken only data from 2022-2023, and only the first admission for each patient. We have excluded admissions where the person died on the day of admission.

We have fitted multivariate Cox proportional hazard models to our data to assess how various factors independently affect the risk of readmission occurring over time.

Assumptions of Cox proportional hazards models

Cox proportional hazards models assume that the hazard rate (in this case risk of death) for a person with one set of covariates is proportional to the hazard rate for a person with a different set of covariates, so that the hazard ratio is assumed to be constant over time.

The proportional hazards (PH) assumption can be checked using statistical tests and graphical diagnostics based on the scaled Schoenfeld residuals.

We find significant p-values for some covariates, particularly for ICBs, indicating that the data violates the proportional hazards assumption. This is unsurprising as we have very high number of patients included within our dataset, which means even very small deviations in the residuals over time can result in significant p-values.

Plotting a smoothed fit to the Schoenfeld residuals allows us to determine, where p-values are significant, whether the changes in the beta coefficient over time are large enough to meaningfully impact our analyses. Deviations from a horizontal line indicate non-proportional hazards, and that the hazard ratio will be changing over time.

There are alternatives when the proportional hazards assumption is violated such as adding a time interaction into the model or using an alternative accelerated failure model. With the size of our dataset and the number of covariates (over 40 ICBs) the alternatives do not appear feasible and would unlikely add useful information to our conclusions.

Considering these factors and that for this analysis the degree of variation and trends are more important than the absolute hazard ratios we have used a Cox proportional hazards models, and the calculated hazard ratios can be considered as a time-averaged hazard ratio. We have modelled the risk of death within 28 days and within 1 yr. Considering the first 28 day period alongside the 1 year time period means we can check for any initial differences in the first 28 days that might be averaged out in the long 1 year period. There are some differences in the hazard ratios calculated for the two models, but the general trends are consistent.

Elderly emergency

Forest plot of hazard ratios. Black indicates no significant difference compared to the reference group, red indicates a significant increase in the hazard ratio (risk) and green indicates a significant decrease. The size of the point gives an indication of the relative number of people in each group.

              chisq df      p sig
sex         41.2493  1 0.0000   *
age_range  150.1131  1 0.0000   *
ethnicity    0.2715  1 0.6023    
imd_decile   0.0999  1 0.7520    
icb          1.3890  1 0.2386    
GLOBAL     243.3581 59 0.0000   *

Forest plot of hazard ratios. Black indicates no significant difference compared to the reference group, red indicates a significant increase in the hazard ratio (risk) and green indicates a significant decrease. The size of the point gives an indication of the relative number of people in each group.

              chisq df      p sig
sex         38.1467  1 0.0000   *
age_range  396.9170  1 0.0000   *
ethnicity    0.0115  1 0.9146    
imd_decile   1.6895  1 0.1937    
icb          1.5642  1 0.2111    
GLOBAL     689.5564 59 0.0000   *

Summary

In the elderly emergency cohort the risk of death post-admission:

  • is higher in males.

  • increases with increasing age.

  • up to 28 days those who are Black/Black British or White Other ethnicity have a lower risk of death compared to White British, while Other and Unknown ethnicities have an increased risk. In the 1 year post-admission the risk of death is lower compared to White British for all other ethnicities, apart from Unknown.

  • increases with increasing deprivation, those living in the most deprived areas (IMD decile 1) have a higher risk of death than those living in the least deprived areas (IMD decile 10).

  • varies by ICB area. The risk of death is lowest in the North West London ICB area and highest in the Norfolk and Waveney ICB area.

Falls

Forest plot of hazard ratios. Black indicates no significant difference compared to the reference group, red indicates a significant increase in the hazard ratio (risk) and green indicates a significant decrease. The size of the point gives an indication of the relative number of people in each group.

              chisq df      p sig
sex          0.0269  1 0.8696    
age_range  400.3599  1 0.0000   *
ethnicity    0.3301  1 0.5656    
imd_decile   0.0387  1 0.8441    
icb         96.5997  1 0.0000   *
GLOBAL      87.4781 61 0.0148   *

Forest plot of hazard ratios. Black indicates no significant difference compared to the reference group, red indicates a significant increase in the hazard ratio (risk) and green indicates a significant decrease. The size of the point gives an indication of the relative number of people in each group.

               chisq df      p sig
sex           0.3263  1 0.5678    
age_range  1138.7820  1 0.0000   *
ethnicity     0.0002  1 0.9881    
imd_decile    0.0035  1 0.9526    
icb           1.3590  1 0.2437    
GLOBAL      213.6630 61 0.0000   *

Summary

In the falls cohort the risk of death post-admission:

  • is higher in males.

  • increases with increasing age.

  • up to 28 days those who are Asian/Asian British, Black/Black British, Other or Unknown ethnicity have a lower risk of death compared to White British. In the year post-admission the risk of death is lower for those who are Black/Black British or White Other ethnicities compared to White British.

  • up to 28days is not as clearly correlated with deprivation. However, in the year post-admission risk of death increases with increasing deprivation, those living in the most deprived areas (IMD decile 1) have a higher risk of death than those living in the least deprived areas (IMD decile 10).

  • varies by ICB area. The risk of death in the year post-admission is lowest in the North West London ICB and North Central London ICB areas, and highest in the Gloucestershire ICB and Mind and South Essex ICB areas. In the 28 days post-admission the risk of readmission is lowest in the South East London ICB area, and highest in the Gloucestershire ICB area.

Frail

Forest plot of hazard ratios. Black indicates no significant difference compared to the reference group, red indicates a significant increase in the hazard ratio (risk) and green indicates a significant decrease. The size of the point gives an indication of the relative number of people in each group.

               chisq df      p sig
sex          22.3987  1 0.0000   *
age_range  5932.7015  1 0.0000   *
ethnicity     0.0645  1 0.7995    
imd_decile    0.6659  1 0.4145    
icb           0.9446  1 0.3311    
GLOBAL      345.5420 61 0.0000   *

Forest plot of hazard ratios. Black indicates no significant difference compared to the reference group, red indicates a significant increase in the hazard ratio (risk) and green indicates a significant decrease. The size of the point gives an indication of the relative number of people in each group.

               chisq df      p sig
sex          23.8176  1 0.0000   *
age_range  3445.9465  1 0.0000   *
ethnicity     0.2245  1 0.6356    
imd_decile    0.0106  1 0.9178    
icb           1.5006  1 0.2206    
GLOBAL      528.0278 61 0.0000   *

Summary

In the frail cohort the risk of death post-admission:

  • is higher in males.

  • increases with increasing age.

  • up to 28 days those who are of Unknown ethnicity have a higher risk of death compared to White British. While in the 1 year post-admission the risk of death is lower for those of Asian/Asian British, Black/Black British and Other ethnicities compared to White British.

  • increases with increasing deprivation, those living in the most deprived areas (IMD decile 1) have a higher risk of death than those living in the least deprived areas (IMD decile 10).

  • varies by ICB area. The risk of death in the year post-admission is lowest in the North West London ICB and Dorset ICB areas and highest in the Humber and North Yorkshire ICB and Hereford and Worcestershire ICB areas. In the 28 days post-admission the risk of readmission is lowest in the North West London ICB and Bristol, North Somerset and South Gloucestershire ICB areas, and highest in the Hereford and Worcestershire ICB and Cornwall and Isles of Scilly ICB areas.

Number of deaths on day of admission by ethnicity

For the survival analysis we excluded anyone who died on the day of first admission, so given the differences seen in the risk of death by ethnicity we have checked whether there are differences in the percentage of patients who die on the day of first admission by ethnicity.

Ethnicity

Percentage who die on day of admissions

White British

0.33

Asian/Asian British

0.33

Black/Black British

0.33

Mixed

0.30

Other

0.25

White Other

0.27

Unknown

0.44

The percentage of patient who die on the day of admission is slightly higher for those who are Asian/Asian British and of Unknown ethnicity compared to those who are White British.

Readmissions

Summary of percentage of patients readmitted within 1 year (2022/23)

All admissions are counted an individual events, so some patients with multiple admissions will be included more than once.

Elderly Emergency

Frail

Falls

Sex

female

275,117 (30.6%)

166,668 (29.4%)

21,796 (21.7%)

male

277,615 (36.4%)

173,647 (33.9%)

15,633 (31.2%)

Age

65-69

N/A

22,486 (18.8%)

966 (9.8%)

70-74

N/A

35,020 (22.3%)

2,048 (13.2%)

75-79

122,350 (24.8%)

53,864 (26%)

4,190 (17.1%)

80-84

136,797 (29.8%)

66,343 (31%)

6,819 (22.3%)

85-89

148,913 (36.7%)

78,824 (37.8%)

10,065 (28.1%)

90+

144,672 (47.6%)

83,778 (48.6%)

13,341 (38.8%)

Ethnicity

Asian/Asian British

12,444 (28.2%)

8,031 (25.3%)

526 (19.5%)

Black/Black British

6,225 (29.6%)

3,872 (27.5%)

252 (19.4%)

Mixed

1,239 (29.8%)

858 (28.5%)

73 (22.3%)

Other

5,554 (30.3%)

3,503 (26.8%)

354 (20.3%)

White British

459,853 (33.7%)

281,826 (32.1%)

31,434 (25.4%)

White Other

18,946 (32.4%)

11,678 (30.6%)

1,280 (23%)

Unknown

48,471 (32.4%)

30,547 (29.9%)

3,510 (22.9%)

IMD decile

1

49,177 (35.7%)

32,826 (32.3%)

3,116 (25%)

2

51,303 (35.2%)

33,762 (32.9%)

3,582 (26.1%)

3

52,017 (34.5%)

33,125 (32.2%)

3,558 (25.5%)

4

55,425 (34.3%)

34,205 (32.2%)

3,711 (25.4%)

5

57,084 (33.4%)

34,999 (31.8%)

3,796 (24.7%)

6

59,006 (33%)

35,802 (31.4%)

4,070 (24.9%)

7

59,835 (32.7%)

35,850 (31.3%)

4,050 (24.8%)

8

59,305 (32.4%)

35,482 (31.2%)

3,972 (24.3%)

9

56,965 (31.6%)

33,744 (30.3%)

4,007 (24.7%)

10

52,615 (31%)

30,520 (29.8%)

3,567 (23.4%)

Total

552,732 (33.3%)

340,315 (31.5%)

37,429 (24.8%)

Number of readmissions in 2022/2023

cohort

mean

median

min

lower quartile

upper quartile

max

elderly emergency

0.55

0

0

0

1

67

falls

0.08

0

0

0

0

7

frail

0.31

0

0

0

0

36

The majority of patients have no readmission within a year, and of those that do have a readmission it is most common to just have a single readmission. A small number of patients are having more regular emergency admissions; those with a highest rates appear to be often admitted and discharged on the same day.

Comparing time to readmission between the cohorts using Kaplan-meier plots (2022/23)

Survival analysis was conducted where each admission is treated as an independent event. This means some patients who had multiple admissions within 2022/23 will be included multiple times.

Survival analysis was conducted using a patient’s first admission within 2022/23, so only one event is included per patient, even if the patient has multiple admissions over the year.

The risk of readmission is very similar between the elderly emergency and frail cohorts (note the frail cohort is a subset of the elderly emergency cohort). Those in the falls cohort have a lower risk of readmission.

The risk of admissions is increased slightly by a small subset of patients that have multiple readmissions, as the risk of readmission decreases if only the first admission/ readmission is included for each patient.

Kaplan Meier survival plots of time to readmission (2022/23)

For these plots each admission in 2022/23 was treated as an independent event, meaning that some patients who had multiple admissions within 2022/23 will be included multiple times.

Summary

The risk of readmission is higher in those who are:

  • aged 80-89, compared to those aged 75-79 or 90+.

  • male

  • Asian/Asian British or Black/Black British compared to White British

  • living in a more deprived area

The risk of readmission varies by ICB area, with a lower risk of readmission in the Dorset, Gloucestershire and Cornwall and Isles of Scilly ICB areas and a higher risk in the North West London ICB area.


ICB

Readmissions

No. at risk

Percentage

QR1: NHS Gloucestershire ICB

7,538

17,057

44.2

QT6: NHS Cornwall & Isles Of Scilly ICB

6,612

14,681

45.0

QVV: NHS Dorset ICB

14,623

31,973

45.7

QOC: NHS Shrops, Telford & Wrekin ICB

7,782

16,293

47.8

QGH: NHS Herefords & Worcestershire ICB

11,357

23,706

47.9

QMM: NHS Norfolk And Waveney ICB

16,713

34,847

48.0

QJM: NHS Lincolnshire ICB

12,130

25,241

48.1

QOX: NHS Bath & NE S'set, S'don & W ICB

13,053

27,036

48.3

QJG: NHS Suffolk & North East Essex ICB

16,829

34,195

49.2

QNX: NHS Sussex ICB

28,798

58,172

49.5

QJK: NHS Devon ICB

21,323

42,920

49.7

QE1: NHS Lancashire & South Cumbria ICB

27,931

55,937

49.9

QSL: NHS Somerset ICB

11,022

22,076

49.9

QKS: NHS Kent And Medway ICB

28,752

57,557

50.0

QK1: NHS Leics, Leic'shire & Rutland ICB

14,608

28,846

50.6

QOQ: NHS Humber And North Yorkshire ICB

30,373

59,446

51.1

QT1: NHS Nott'ham & Nottinghamshire ICB

16,915

32,881

51.4

QWO: NHS West Yorkshire ICB

32,617

63,464

51.4

QU9: NHS Bucks, Oxford & Berkshire W ICB

24,114

46,592

51.8

QH8: NHS Mid And South Essex ICB

20,108

38,759

51.9

QM7: NHS Hertfordshire & West Essex ICB

22,898

44,110

51.9

QUY: NHS Bristol, N S'set & S Gloucs ICB

14,302

27,504

52.0

QYG: NHS Cheshire And Merseyside ICB

43,588

83,759

52.0

QOP: NHS Greater Manchester ICB

40,329

76,617

52.6

QMJ: NHS North Central London ICB

12,730

24,146

52.7

QHG: NHS Beds, Luton & Milton Keynes ICB

13,493

25,501

52.9

QRL: NHS Hampshire And Isle Of Wight ICB

33,940

64,205

52.9

QWE: NHS South West London ICB

16,393

30,997

52.9

QWU: NHS Coventry And Warwickshire ICB

16,336

30,824

53.0

QXU: NHS Surrey Heartlands ICB

18,816

35,481

53.0

QUE: NHS Cambs & Peterborough ICB

14,133

26,315

53.7

QJ2: NHS Derby And Derbyshire ICB

20,582

37,959

54.2

QUA: NHS Black Country ICB

22,677

41,709

54.4

QHM: NHS North East & North Cumbria ICB

62,334

114,454

54.5

QNC: NHS Staffordshire & Stoke-On-Tr ICB

22,617

41,333

54.7

QHL: NHS Birmingham And Solihull ICB

21,020

38,199

55.0

QMF: NHS North East London ICB

18,429

32,989

55.9

QF7: NHS South Yorkshire ICB

26,758

47,738

56.1

QKK: NHS South East London ICB

21,158

37,627

56.2

QPM: NHS Northamptonshire ICB

13,206

23,500

56.2

QRV: NHS North West London ICB

28,710

49,403

58.1

Summary

The risk of readmission is higher in those who are:

  • aged over 75 years.

  • male

  • Black/Black British

In this falls cohort there is no clear association between deprivation and risk of readmission.

The risk of readmission varies by ICB area.


ICB

Readmissions

No. at risk

Percentage

QT6: NHS Cornwall & Isles Of Scilly ICB

443

1,252

35.4

QR1: NHS Gloucestershire ICB

572

1,606

35.6

QOC: NHS Shrops, Telford & Wrekin ICB

518

1,344

38.5

QGH: NHS Herefords & Worcestershire ICB

925

2,311

40.0

QJK: NHS Devon ICB

1,380

3,351

41.2

QJM: NHS Lincolnshire ICB

914

2,213

41.3

QK1: NHS Leics, Leic'shire & Rutland ICB

1,024

2,435

42.1

QSL: NHS Somerset ICB

857

1,996

42.9

QMM: NHS Norfolk And Waveney ICB

1,546

3,599

43.0

QOX: NHS Bath & NE S'set, S'don & W ICB

1,169

2,677

43.7

QE1: NHS Lancashire & South Cumbria ICB

2,316

5,263

44.0

QOQ: NHS Humber And North Yorkshire ICB

2,061

4,683

44.0

QVV: NHS Dorset ICB

1,279

2,874

44.5

QJG: NHS Suffolk & North East Essex ICB

1,420

3,170

44.8

QNX: NHS Sussex ICB

2,675

5,928

45.1

QJ2: NHS Derby And Derbyshire ICB

1,171

2,593

45.2

QT1: NHS Nott'ham & Nottinghamshire ICB

1,293

2,831

45.7

QU9: NHS Bucks, Oxford & Berkshire W ICB

1,857

4,040

46.0

QWO: NHS West Yorkshire ICB

2,759

5,867

47.0

QUY: NHS Bristol, N S'set & S Gloucs ICB

1,333

2,800

47.6

QYG: NHS Cheshire And Merseyside ICB

3,414

7,141

47.8

QMJ: NHS North Central London ICB

1,358

2,785

48.8

QKS: NHS Kent And Medway ICB

1,790

3,642

49.1

QOP: NHS Greater Manchester ICB

3,352

6,824

49.1

QM7: NHS Hertfordshire & West Essex ICB

2,077

4,211

49.3

QHM: NHS North East & North Cumbria ICB

4,772

9,638

49.5

QXU: NHS Surrey Heartlands ICB

1,555

3,132

49.6

QUE: NHS Cambs & Peterborough ICB

1,243

2,462

50.5

QHG: NHS Beds, Luton & Milton Keynes ICB

1,363

2,681

50.8

QWU: NHS Coventry And Warwickshire ICB

1,413

2,776

50.9

QH8: NHS Mid And South Essex ICB

2,050

4,017

51.0

QF7: NHS South Yorkshire ICB

2,098

4,060

51.7

QNC: NHS Staffordshire & Stoke-On-Tr ICB

1,523

2,935

51.9

QRL: NHS Hampshire And Isle Of Wight ICB

3,505

6,745

52.0

QUA: NHS Black Country ICB

2,004

3,857

52.0

QWE: NHS South West London ICB

1,666

3,204

52.0

QHL: NHS Birmingham And Solihull ICB

1,585

3,007

52.7

QPM: NHS Northamptonshire ICB

1,085

2,016

53.8

QMF: NHS North East London ICB

2,087

3,864

54.0

QRV: NHS North West London ICB

2,484

4,569

54.4

QKK: NHS South East London ICB

2,702

4,756

56.8

Summary

The risk of readmission is higher in those who are:

  • aged 75-89 years, with those aged under 75 having the lowest risk of readmission

  • male

  • Black/Black British

  • living in the most deprived areas

The risk of readmission varies by ICB area, with a lower risk of readmission in the Gloucestershire and Cornwall and Isles of Scilly ICB areas and a higher risk in the North West London ICB area.


ICB

Readmissions

No. at risk

Percentage

QR1: NHS Gloucestershire ICB

4,229

10,449

40.5

QT6: NHS Cornwall & Isles Of Scilly ICB

3,715

9,002

41.3

QVV: NHS Dorset ICB

8,213

19,215

42.7

QOC: NHS Shrops, Telford & Wrekin ICB

4,528

10,308

43.9

QJM: NHS Lincolnshire ICB

7,508

16,864

44.5

QGH: NHS Herefords & Worcestershire ICB

6,758

15,078

44.8

QMM: NHS Norfolk And Waveney ICB

9,940

22,141

44.9

QOX: NHS Bath & NE S'set, S'don & W ICB

7,643

16,719

45.7

QJK: NHS Devon ICB

11,802

25,526

46.2

QSL: NHS Somerset ICB

6,411

13,818

46.4

QE1: NHS Lancashire & South Cumbria ICB

17,297

37,117

46.6

QJG: NHS Suffolk & North East Essex ICB

9,961

21,267

46.8

QK1: NHS Leics, Leic'shire & Rutland ICB

8,621

18,422

46.8

QNX: NHS Sussex ICB

17,634

37,642

46.8

QKS: NHS Kent And Medway ICB

15,246

32,496

46.9

QOQ: NHS Humber And North Yorkshire ICB

17,918

37,982

47.2

QT1: NHS Nott'ham & Nottinghamshire ICB

10,601

21,955

48.3

QH8: NHS Mid And South Essex ICB

12,407

25,602

48.5

QU9: NHS Bucks, Oxford & Berkshire W ICB

14,518

29,853

48.6

QWO: NHS West Yorkshire ICB

21,157

43,545

48.6

QHG: NHS Beds, Luton & Milton Keynes ICB

7,893

16,223

48.7

QM7: NHS Hertfordshire & West Essex ICB

13,667

27,985

48.8

QOP: NHS Greater Manchester ICB

25,408

51,928

48.9

QUY: NHS Bristol, N S'set & S Gloucs ICB

8,593

17,524

49.0

QYG: NHS Cheshire And Merseyside ICB

28,657

58,494

49.0

QMJ: NHS North Central London ICB

7,837

15,782

49.7

QWE: NHS South West London ICB

10,510

21,047

49.9

QWU: NHS Coventry And Warwickshire ICB

10,095

20,184

50.0

QXU: NHS Surrey Heartlands ICB

11,267

22,544

50.0

QHM: NHS North East & North Cumbria ICB

38,245

75,965

50.3

QRL: NHS Hampshire And Isle Of Wight ICB

20,006

39,710

50.4

QUA: NHS Black Country ICB

14,083

27,911

50.5

QUE: NHS Cambs & Peterborough ICB

8,632

17,067

50.6

QHL: NHS Birmingham And Solihull ICB

12,592

24,665

51.1

QNC: NHS Staffordshire & Stoke-On-Tr ICB

14,103

27,412

51.4

QJ2: NHS Derby And Derbyshire ICB

12,601

24,212

52.0

QF7: NHS South Yorkshire ICB

15,998

30,700

52.1

QMF: NHS North East London ICB

11,630

22,205

52.4

QKK: NHS South East London ICB

14,170

26,579

53.3

QRV: NHS North West London ICB

17,882

33,377

53.6

QPM: NHS Northamptonshire ICB

8,465

15,729

53.8

Multivariate survival analysis of time to readmission (2022/23)

For this analysis we have taken only data from 2022-2023, and only the first admission/time to readmission for each patient.

We have fitted Cox proportional hazard models to our data to assess how various factors independently affect the risk of readmission occurring over time.

For information on the testing of the proportional hazards assumption see Section 1.5.1.

We have used a Cox proportional hazards models and the calculated hazard ratios can be considered as a time-averaged hazard ratio. We have considered models for readmission within 28 days and within 1 yr. There are some differences between the hazard ratios calculated for the two models, but the general trends are consistent.

Elderly emergency

Forest plot of hazard ratios. Black indicates no significant difference compared to the reference group, red indicates a significant increase in the hazard ratio (risk) and green indicates a significant decrease. The size of the point gives an indication of the relative number of people in each group.

              chisq df      p sig
sex          0.5985  1 0.4391    
age_range    0.0000  1 0.9997    
ethnicity    0.3326  1 0.5641    
imd_decile   0.1206  1 0.7284    
icb         26.2954  1 0.0000   *
GLOBAL     648.1661 59 0.0000   *

Forest plot of hazard ratios. Black indicates no significant difference compared to the reference group, red indicates a significant increase in the hazard ratio (risk) and green indicates a significant decrease. The size of the point gives an indication of the relative number of people in each group.

               chisq df      p sig
sex         286.7574  1 0.0000   *
age_range     0.3894  1 0.5326    
ethnicity     0.2177  1 0.6408    
imd_decile    2.6653  1 0.1026    
icb           3.9941  1 0.0457   *
GLOBAL     1147.1447 59 0.0000   *

Summary

In the elderly emergency cohort the risk of readmission:

  • is higher in males.

  • is higher in those aged 80-89 years compared to those aged 75-79 years. Up to 28 days the risk of readmission in those aged 90+ is no higher than those aged 75-79 years, but in the full year after admission the risk of readmission is higher in those over 90 years than those aged 75-79 years. This may be influenced by the risk of death being significantly different between those aged 75-79 years and those aged over 90.

  • is higher in those who are White British, and significantly lower in those who are Asian/Asian British, Black/Black British, Other or Unknown ethnicity. Up to 28 days those of mixed ethnicity have a similar risk of readmission to White British, but in the analysis, but the risk drops below that of White British in the analysis to 1 year.

  • increases with increasing deprivation, those living in the most deprived areas (IMD decile 1) are more likely to be readmitted than those living in the least deprived areas (IMD decile 10).

  • varies by ICB area. The risk of readmission is lowest in the Cornwall and Isles of Scilly ICB area. In the year post-admission the risk of readmission is highest in South East London ICB and North West London ICB areas. While up to 28 days post-admission the risk is highest in the North East London ICB area.

Falls

Forest plot of hazard ratios. Black indicates no significant difference compared to the reference group, red indicates a significant increase in the hazard ratio (risk) and green indicates a significant decrease. The size of the point gives an indication of the relative number of people in each group.

              chisq df      p sig
sex          3.2766  1 0.0703    
age_range   34.8998  1 0.0000   *
ethnicity    0.0642  1 0.8000    
imd_decile   0.3033  1 0.5818    
icb         13.2508  1 0.0003   *
GLOBAL     115.2325 61 0.0000   *

Forest plot of hazard ratios. Black indicates no significant difference compared to the reference group, red indicates a significant increase in the hazard ratio (risk) and green indicates a significant decrease. The size of the point gives an indication of the relative number of people in each group.

              chisq df      p sig
sex          7.3360  1 0.0068   *
age_range   16.1494  1 0.0001   *
ethnicity    0.0009  1 0.9757    
imd_decile   0.0803  1 0.7769    
icb         18.6529  1 0.0000   *
GLOBAL     109.3210 61 0.0001   *

Summary

In the falls cohort the risk of readmission:

  • is higher in males.

  • is higher in those aged over 70 years compared to those aged 60-64 years.

  • is lower in those of Unknown ethnicity compared to White British. In the analysis to 1 year the risk of readmission is significantly higher in those who are Black/Black British compared to those who are White British.

  • increases with increasing deprivation, those living in the most deprived areas (IMD decile 1) are more likely to be readmitted than those living in the least deprived areas (IMD decile 10).

  • varies by ICB area. The risk of readmission in the year post-admission is lowest in the Cornwall and Isles of Scilly ICB and Gloucestershire ICB areas and highest in the South East London ICB and Northamptonshire ICB areas. In the 28 days post-admission the risk of readmission is lowest in the Gloucestershire ICB and Hereford and Worcestershire ICB areas, and highest in the North West London ICB and Staffordshire and Stoke-on-Trent ICB areas.

Frail

Forest plot of hazard ratios. Black indicates no significant difference compared to the reference group, red indicates a significant increase in the hazard ratio (risk) and green indicates a significant decrease. The size of the point gives an indication of the relative number of people in each group.

              chisq df      p sig
sex          1.5728  1 0.2098    
age_range    1.7869  1 0.1813    
ethnicity    0.3570  1 0.5502    
imd_decile   0.0945  1 0.7585    
icb        104.4060  1 0.0000   *
GLOBAL     532.4705 61 0.0000   *

Forest plot of hazard ratios. Black indicates no significant difference compared to the reference group, red indicates a significant increase in the hazard ratio (risk) and green indicates a significant decrease. The size of the point gives an indication of the relative number of people in each group.

               chisq df      p sig
sex         158.5625  1 0.0000   *
age_range   436.9416  1 0.0000   *
ethnicity     0.2432  1 0.6219    
imd_decile    3.4745  1 0.0623    
icb           4.6099  1 0.0318   *
GLOBAL     1178.8713 61 0.0000   *

Summary

In the frail cohort the risk of readmission:

  • is higher in males.

  • up to 28days age does not significantly influence the risk of readmission.However, in the year post admission the risk of readmission is higher in those aged over 80 years compared to those aged 75-79 years.

  • is higher in those who are White British, and significantly lower in those who are Asian/Asian British, Other and Unknown ethnicity. Up to 28 days those who are Black/Black British also have a lower risk of readmission compared to White British, but the risk is no longer reduced in the analysis to 1 year.

  • increases with increasing deprivation, those living in the most deprived areas (IMD decile 1) are more likely to be readmitted than those living in the least deprived areas (IMD decile 10).

  • varies by ICB area. The risk of readmission in the year post-admission is lowest in the Cornwall and Isles of Scilly ICB and Gloucestershire ICB areas and highest in the South East London ICB and North West London ICB areas. In the 28 days post-admission the risk of readmission is lowest in the Cornwall and Isles of Scilly ICB and Leicester, Leicestershire and Rutland ICB areas, and highest in the Staffordshire and Stoke-on-Trent ICB and North East London ICB areas.

Readmissions- excluding those who died prior to readmission

In the previous analysis of readmissions we see that some of the ICB areas with the lowest readmission rates are those with higher mortality rates. Given that the risk of readmission will be strongly influenced by the mortality rate for this analysis we have excluded all those that died prior to readmission at either 28 days or 1 year. This gives us the risk of readmission for those that survived to the end of the time period of interest, either 28 days or 1 year.

Summary of percentage of patients readmitted within 1 year (2022/23)

All admissions are counted an individual events, so some patients with multiple admissions will be included more than once.

Elderly Emergency

Frail

Falls

Sex

female

137,552 (18%)

81,461 (16.9%)

11,819 (13%)

male

150,783 (23.7%)

92,944 (21.6%)

9,353 (21.4%)

Age

65-69

N/A

13,823 (12.4%)

665 (7%)

70-74

N/A

20,484 (14.4%)

1,330 (9%)

75-79

70,759 (16%)

30,062 (16.4%)

2,612 (11.4%)

80-84

74,955 (18.9%)

34,796 (19.1%)

4,115 (14.8%)

85-89

76,842 (23%)

38,909 (23.1%)

5,798 (18.4%)

90+

65,779 (29.2%)

36,331 (29.1%)

6,652 (24%)

Ethnicity

Asian/Asian British

7,018 (18.2%)

4,431 (15.7%)

351 (13.9%)

Black/Black British

3,524 (19.2%)

2,179 (17.6%)

178 (14.5%)

Mixed

669 (18.7%)

479 (18.2%)

46 (15.3%)

Other

2,802 (18%)

1,717 (15.2%)

203 (12.8%)

White British

241,639 (21.1%)

145,474 (19.6%)

17,898 (16.2%)

White Other

10,243 (20.6%)

6,141 (18.8%)

773 (15.3%)

Unknown

22,440 (18.1%)

13,984 (16.3%)

1,723 (12.7%)

IMD decile

1

26,455 (23%)

17,160 (20%)

1,834 (16.4%)

2

27,277 (22.4%)

17,521 (20.3%)

2,145 (17.5%)

3

27,735 (21.9%)

17,417 (20%)

2,126 (17%)

4

28,717 (21.3%)

17,295 (19.4%)

2,134 (16.4%)

5

29,722 (20.7%)

17,854 (19.3%)

2,088 (15.3%)

6

30,450 (20.2%)

18,229 (18.9%)

2,263 (15.5%)

7

30,769 (20%)

18,049 (18.7%)

2,203 (15.2%)

8

30,652 (19.9%)

18,040 (18.7%)

2,245 (15.3%)

9

29,339 (19.2%)

17,166 (18.1%)

2,151 (15%)

10

27,219 (18.9%)

15,674 (17.9%)

1,983 (14.5%)

Total

288,335 (20.6%)

174,405 (19.1%)

21,172 (15.8%)

Comparing time to readmission between the cohorts using Kaplan-meier plots (2022/23)

Survival analysis was conducted where each admission is treated as an independent event. This means some patients who had multiple admissions within 2022/23 will be included multiple times.

Survival analysis was conducted using a patient’s first admission within 2022/23, so only one event is included per patient, even if the patient has multiple admissions over the year.

The risk of readmission is very similar between the elderly emergency and frail cohorts (note the frail cohort is a subset of the elderly emergency cohort). Those in the falls cohort have a lower risk of readmission.

The risk of admissions is increased slightly by a small subset of patients that have multiple readmissions, as the risk of readmission decreases if only the first admission/ readmission is included for each patient.

Kaplan Meier survival plots of time to readmission (2022/23)

For these plots each admission in 2022/23 was treated as an independent event, meaning that some patients who had multiple admissions within 2022/23 will be included multiple times.

Summary

The risk of readmission is higher in those who are:

  • older

  • male

  • Black/Black British

  • living in a more deprived area

The risk of readmission varies by ICB area, with a lower risk of readmission in the Dorset, Gloucestershire and Cornwall and Isles of Scilly ICB areas.


ICB

Readmissions

No. at risk

Percentage

QR1: NHS Gloucestershire ICB

7,536

14,085

53.5

QT6: NHS Cornwall & Isles Of Scilly ICB

6,608

12,168

54.3

QVV: NHS Dorset ICB

14,615

26,857

54.4

QOC: NHS Shrops, Telford & Wrekin ICB

7,780

13,660

57.0

QOX: NHS Bath & NE S'set, S'don & W ICB

13,049

22,715

57.4

QGH: NHS Herefords & Worcestershire ICB

11,353

19,501

58.2

QJM: NHS Lincolnshire ICB

12,121

20,815

58.2

QMM: NHS Norfolk And Waveney ICB

16,707

28,421

58.8

QJK: NHS Devon ICB

21,309

35,933

59.3

QKS: NHS Kent And Medway ICB

28,737

48,345

59.4

QJG: NHS Suffolk & North East Essex ICB

16,823

28,277

59.5

QNX: NHS Sussex ICB

28,783

48,358

59.5

QSL: NHS Somerset ICB

11,011

18,439

59.7

QE1: NHS Lancashire & South Cumbria ICB

27,919

46,422

60.1

QK1: NHS Leics, Leic'shire & Rutland ICB

14,600

24,309

60.1

QU9: NHS Bucks, Oxford & Berkshire W ICB

24,095

39,685

60.7

QUY: NHS Bristol, N S'set & S Gloucs ICB

14,292

23,364

61.2

QOQ: NHS Humber And North Yorkshire ICB

30,354

49,452

61.4

QRL: NHS Hampshire And Isle Of Wight ICB

33,922

54,811

61.9

QMJ: NHS North Central London ICB

12,725

20,531

62.0

QM7: NHS Hertfordshire & West Essex ICB

22,891

36,862

62.1

QT1: NHS Nott'ham & Nottinghamshire ICB

16,909

27,218

62.1

QH8: NHS Mid And South Essex ICB

20,102

32,310

62.2

QYG: NHS Cheshire And Merseyside ICB

43,565

70,074

62.2

QXU: NHS Surrey Heartlands ICB

18,800

30,174

62.3

QWO: NHS West Yorkshire ICB

32,610

52,301

62.4

QWE: NHS South West London ICB

16,387

26,186

62.6

QWU: NHS Coventry And Warwickshire ICB

16,321

25,980

62.8

QHG: NHS Beds, Luton & Milton Keynes ICB

13,487

21,456

62.9

QOP: NHS Greater Manchester ICB

40,307

63,630

63.3

QNC: NHS Staffordshire & Stoke-On-Tr ICB

22,610

35,316

64.0

QHM: NHS North East & North Cumbria ICB

62,301

96,995

64.2

QJ2: NHS Derby And Derbyshire ICB

20,574

32,050

64.2

QUE: NHS Cambs & Peterborough ICB

14,128

22,021

64.2

QUA: NHS Black Country ICB

22,664

35,223

64.3

QHL: NHS Birmingham And Solihull ICB

21,012

32,608

64.4

QMF: NHS North East London ICB

18,425

28,158

65.4

QKK: NHS South East London ICB

21,149

32,186

65.7

QRV: NHS North West London ICB

28,701

43,514

66.0

QPM: NHS Northamptonshire ICB

13,202

19,968

66.1

QF7: NHS South Yorkshire ICB

26,747

40,369

66.3

Summary

The risk of readmission is higher in those who are:

  • older

  • male

  • Black/Black British

  • living in a more deprived area

The risk of readmission varies by ICB area.


ICB

Readmissions

No. at risk

Percentage

QT6: NHS Cornwall & Isles Of Scilly ICB

443

1,099

40.3

QR1: NHS Gloucestershire ICB

572

1,326

43.1

QOC: NHS Shrops, Telford & Wrekin ICB

518

1,180

43.9

QGH: NHS Herefords & Worcestershire ICB

925

2,005

46.1

QJK: NHS Devon ICB

1,380

2,925

47.2

QJM: NHS Lincolnshire ICB

914

1,928

47.4

QK1: NHS Leics, Leic'shire & Rutland ICB

1,024

2,117

48.4

QSL: NHS Somerset ICB

857

1,752

48.9

QOX: NHS Bath & NE S'set, S'don & W ICB

1,169

2,381

49.1

QMM: NHS Norfolk And Waveney ICB

1,546

3,116

49.6

QE1: NHS Lancashire & South Cumbria ICB

2,316

4,650

49.8

QVV: NHS Dorset ICB

1,279

2,565

49.9

QOQ: NHS Humber And North Yorkshire ICB

2,058

4,090

50.3

QJG: NHS Suffolk & North East Essex ICB

1,420

2,808

50.6

QJ2: NHS Derby And Derbyshire ICB

1,170

2,297

50.9

QT1: NHS Nott'ham & Nottinghamshire ICB

1,293

2,534

51.0

QU9: NHS Bucks, Oxford & Berkshire W ICB

1,857

3,638

51.0

QNX: NHS Sussex ICB

2,673

5,229

51.1

QMJ: NHS North Central London ICB

1,357

2,556

53.1

QWO: NHS West Yorkshire ICB

2,757

5,162

53.4

QYG: NHS Cheshire And Merseyside ICB

3,413

6,384

53.5

QKS: NHS Kent And Medway ICB

1,790

3,275

54.7

QM7: NHS Hertfordshire & West Essex ICB

2,077

3,785

54.9

QHM: NHS North East & North Cumbria ICB

4,769

8,655

55.1

QOP: NHS Greater Manchester ICB

3,351

6,069

55.2

QXU: NHS Surrey Heartlands ICB

1,555

2,817

55.2

QUY: NHS Bristol, N S'set & S Gloucs ICB

1,333

2,390

55.8

QUE: NHS Cambs & Peterborough ICB

1,243

2,205

56.4

QWU: NHS Coventry And Warwickshire ICB

1,411

2,492

56.6

QWE: NHS South West London ICB

1,665

2,932

56.8

QHG: NHS Beds, Luton & Milton Keynes ICB

1,363

2,397

56.9

QNC: NHS Staffordshire & Stoke-On-Tr ICB

1,523

2,659

57.3

QRL: NHS Hampshire And Isle Of Wight ICB

3,504

6,113

57.3

QH8: NHS Mid And South Essex ICB

2,050

3,545

57.8

QF7: NHS South Yorkshire ICB

2,098

3,610

58.1

QMF: NHS North East London ICB

2,087

3,568

58.5

QUA: NHS Black Country ICB

2,004

3,428

58.5

QHL: NHS Birmingham And Solihull ICB

1,585

2,703

58.6

QRV: NHS North West London ICB

2,484

4,211

59.0

QPM: NHS Northamptonshire ICB

1,085

1,836

59.1

QKK: NHS South East London ICB

2,701

4,393

61.5

Summary

The risk of readmission is higher in those who are:

  • older

  • male

  • Black/Black British

  • living in the more deprived areas

The risk of readmission varies by ICB area, with a lower risk of readmission in the Dorset, Gloucestershire and Cornwall and Isles of Scilly ICB areas.


ICB

Readmissions

No. at risk

Percentage

QR1: NHS Gloucestershire ICB

4,228

8,632

49.0

QT6: NHS Cornwall & Isles Of Scilly ICB

3,711

7,478

49.6

QVV: NHS Dorset ICB

8,209

16,329

50.3

QOC: NHS Shrops, Telford & Wrekin ICB

4,528

8,677

52.2

QJM: NHS Lincolnshire ICB

7,503

14,022

53.5

QOX: NHS Bath & NE S'set, S'don & W ICB

7,642

14,207

53.8

QGH: NHS Herefords & Worcestershire ICB

6,756

12,403

54.5

QMM: NHS Norfolk And Waveney ICB

9,939

18,153

54.8

QJK: NHS Devon ICB

11,795

21,430

55.0

QSL: NHS Somerset ICB

6,408

11,654

55.0

QK1: NHS Leics, Leic'shire & Rutland ICB

8,620

15,595

55.3

QKS: NHS Kent And Medway ICB

15,240

27,267

55.9

QNX: NHS Sussex ICB

17,627

31,476

56.0

QE1: NHS Lancashire & South Cumbria ICB

17,292

30,821

56.1

QOQ: NHS Humber And North Yorkshire ICB

17,908

31,735

56.4

QJG: NHS Suffolk & North East Essex ICB

9,957

17,584

56.6

QU9: NHS Bucks, Oxford & Berkshire W ICB

14,508

25,628

56.6

QUY: NHS Bristol, N S'set & S Gloucs ICB

8,589

14,929

57.5

QHG: NHS Beds, Luton & Milton Keynes ICB

7,890

13,670

57.7

QH8: NHS Mid And South Essex ICB

12,405

21,477

57.8

QM7: NHS Hertfordshire & West Essex ICB

13,661

23,592

57.9

QT1: NHS Nott'ham & Nottinghamshire ICB

10,598

18,298

57.9

QXU: NHS Surrey Heartlands ICB

11,263

19,424

58.0

QYG: NHS Cheshire And Merseyside ICB

28,643

49,417

58.0

QMJ: NHS North Central London ICB

7,835

13,474

58.1

QWO: NHS West Yorkshire ICB

21,149

36,388

58.1

QOP: NHS Greater Manchester ICB

25,398

43,396

58.5

QWE: NHS South West London ICB

10,505

17,934

58.6

QRL: NHS Hampshire And Isle Of Wight ICB

19,996

34,051

58.7

QHM: NHS North East & North Cumbria ICB

38,226

64,989

58.8

QWU: NHS Coventry And Warwickshire ICB

10,084

17,073

59.1

QHL: NHS Birmingham And Solihull ICB

12,589

21,220

59.3

QUA: NHS Black Country ICB

14,078

23,726

59.3

QNC: NHS Staffordshire & Stoke-On-Tr ICB

14,100

23,730

59.4

QUE: NHS Cambs & Peterborough ICB

8,631

14,278

60.4

QRV: NHS North West London ICB

17,874

29,550

60.5

QJ2: NHS Derby And Derbyshire ICB

12,595

20,681

60.9

QMF: NHS North East London ICB

11,629

19,004

61.2

QF7: NHS South Yorkshire ICB

15,994

26,096

61.3

QKK: NHS South East London ICB

14,163

22,840

62.0

QPM: NHS Northamptonshire ICB

8,464

13,426

63.0

Multivariate survival analysis of time to readmission (2022/23)

For this analysis we have taken only data from 2022-2023, and only the first admission/time to readmission for each patient.

We have fitted Cox proportional hazard models to our data to assess how various factors independently affect the risk of readmission occurring over time.

For information on the testing of the proportional hazards assumption see Section 1.5.1.

We have used a Cox proportional hazards models and the calculated hazard ratios can be considered as a time-averaged hazard ratio. We have considered models for readmission within 28 days and within 1 yr. There are some differences between the hazard ratios calculated for the two models, but the general trends are consistent.

Elderly emergency

Forest plot of hazard ratios. Black indicates no significant difference compared to the reference group, red indicates a significant increase in the hazard ratio (risk) and green indicates a significant decrease. The size of the point gives an indication of the relative number of people in each group.

              chisq df      p sig
sex          0.1925  1 0.6608    
age_range    0.0028  1 0.9580    
ethnicity    0.3260  1 0.5680    
imd_decile   0.3230  1 0.5698    
icb         20.5807  1 0.0000   *
GLOBAL     657.5789 59 0.0000   *

Forest plot of hazard ratios. Black indicates no significant difference compared to the reference group, red indicates a significant increase in the hazard ratio (risk) and green indicates a significant decrease. The size of the point gives an indication of the relative number of people in each group.

               chisq df      p sig
sex         166.1685  1 0.0000   *
age_range    90.2829  1 0.0000   *
ethnicity     0.7149  1 0.3978    
imd_decile   15.9585  1 0.0001   *
icb           2.1304  1 0.1444    
GLOBAL     2028.8874 59 0.0000   *

Summary

In the elderly emergency cohort the risk of readmission:

  • is higher in males.

  • increases with increasing age

  • is higher in those who are White British, and significantly lower in those who are Asian/Asian British, Black/Black British, Other or Unknown ethnicity. Up to 28 days those of mixed ethnicity have a similar risk of readmission to White British, but in the analysis, but the risk drops below that of White British in the analysis to 1 year.

  • increases with increasing deprivation, those living in the most deprived areas (IMD decile 1) are more likely to be readmitted than those living in the least deprived areas (IMD decile 10).

  • varies by ICB area. The risk of readmission is lowest in the Cornwall and Isles of Scilly ICB ares. In the year post-admission the risk of readmission is highest in South East London ICB and Northamptonshire areas. While up to 28 days post-admission the risk is highest in the North East London ICB and South East London ICB areas.

Falls

Forest plot of hazard ratios. Black indicates no significant difference compared to the reference group, red indicates a significant increase in the hazard ratio (risk) and green indicates a significant decrease. The size of the point gives an indication of the relative number of people in each group.

              chisq df      p sig
sex          2.8978  1 0.0887    
age_range   46.2675  1 0.0000   *
ethnicity    0.0725  1 0.7877    
imd_decile   0.3218  1 0.5706    
icb         11.3693  1 0.0007   *
GLOBAL     114.7952 61 0.0000   *

Forest plot of hazard ratios. Black indicates no significant difference compared to the reference group, red indicates a significant increase in the hazard ratio (risk) and green indicates a significant decrease. The size of the point gives an indication of the relative number of people in each group.

              chisq df      p sig
sex          0.5071  1 0.4764    
age_range  580.2587  1 0.0000   *
ethnicity    0.0239  1 0.8770    
imd_decile   0.2608  1 0.6095    
icb          8.2191  1 0.0041   *
GLOBAL     124.2188 61 0.0000   *

Summary

In the falls cohort the risk of readmission:

  • is higher in males.

  • is higher in those aged over 70 years compared to those aged 60-64 years.

  • is lower in those of Unknown ethnicity compared to White British. In the analysis to 1 year post-admission the risk of readmission is also significantly lower in those who are Asian/Asian British or from Other ethnic backgrounds.

  • increases with increasing deprivation, those living in the most deprived areas (IMD decile 1) are more likely to be readmitted than those living in the least deprived areas (IMD decile 10).

  • varies by ICB area. The risk of readmission in the year post-admission is lowest in the Cornwall and Isles of Scilly ICB and Gloucestershire ICB areas and highest in the South East London ICB and Northamptonshire ICB areas. In the 28 days post-admission the risk of readmission is lowest in the Leicester, Leicestershire and Rutland ICB and Hereford and Worcestershire ICB areas, and highest in the North West London ICB and Staffordshire and Stoke-on-Trent ICB areas.

Frail

Forest plot of hazard ratios. Black indicates no significant difference compared to the reference group, red indicates a significant increase in the hazard ratio (risk) and green indicates a significant decrease. The size of the point gives an indication of the relative number of people in each group.

              chisq df      p sig
sex          1.0416  1 0.3074    
age_range    7.0704  1 0.0078   *
ethnicity    0.3923  1 0.5311    
imd_decile   0.2569  1 0.6122    
icb         83.2885  1 0.0000   *
GLOBAL     540.8746 61 0.0000   *

Forest plot of hazard ratios. Black indicates no significant difference compared to the reference group, red indicates a significant increase in the hazard ratio (risk) and green indicates a significant decrease. The size of the point gives an indication of the relative number of people in each group.

               chisq df      p sig
sex         107.2706  1 0.0000   *
age_range  6702.7256  1 0.0000   *
ethnicity     0.8407  1 0.3592    
imd_decile   14.4995  1 0.0001   *
icb           1.6387  1 0.2005    
GLOBAL     2143.9871 61 0.0000   *

Summary

In the frail cohort the risk of readmission:

  • is higher in males.

  • increases with increasing age.

  • is higher in those who are White British, and significantly lower in those who are Asian/Asian British, Black/Black British, Other and Unknown ethnicity. In the analysis to one year post-admission those who are White Other also have a significantly lower risk of readmission.

  • increases with increasing deprivation, those living in the most deprived areas (IMD decile 1) are more likely to be readmitted than those living in the least deprived areas (IMD decile 10).

  • varies by ICB area. The risk of readmission in the year post-admission is lowest in the and Gloucestershire ICB and Cornwall and Isles of Scilly ICB areas and highest in the South East London ICB and Northamptonshire ICB areas. In the 28 days post-admission the risk of readmission is lowest in the Cornwall and Isles of Scilly ICB and Leicester, Leicestershire and Rutland ICB areas, and highest in the Staffordshire and Stoke-on-Trent ICB and North East London ICB areas.

Conclusions

Following an emergency admission the risk of both death, and readmission (in those who do not die within the follow-up period), in all 3 cohorts (elderly emergency, falls or frail) was higher for males, those living in more deprived areas and those who were older at admission.

There was also significant variability in risk of death and readmission between ICB area, even once any differences in the age, ethnicity, sex and deprivation of the areas were accounted for in this analysis.

The risk of death in the year post-admission was lower for those who were Asian/Asian British, Black/Black British or Other ethnicities, compared to White British. There is a general trend towards a higher risk of readmission for those who are White British than those of other ethnicities. In the elderly emergency and frail cohorts the risk of readmission was lower in those who were Asian/Asian British, Black/Black British, Other and Unknown ethnicity. For the falls cohort there are less differences between ethnic groups, but in the 1 year post-admission analysis the risk of readmission is also significantly lower in those who are Asian/Asian British, from Other or Unknown ethnic backgrounds.